New York State Report on Sepsis Care Improvement Initiative: Quality Performance 2015 Office the Medical Director Office Quality and Patient Safety March 2017
Page Intentionally Left Blank
Table Contents Overview... 1 The New York State Sepsis Initiative... 1 Statewide Trends... 2 Sepsis Improvement Initiatives: Collaborations... 7 Measure Descriptions... 9 Performance Data... 11 Next Steps... 23 Technical Appendix A... 25 Technical Appendix B... 29
Page Intentionally Left Blank
Overview Sepsis is defined as a clinical syndrome in which patients have an infection which is accompanied by signs and symptoms a systemic inflammatory response. Sepsis sufficient severity that major organ systems in the body (such as heart, kidney, brain and others) are impaired is referred to as severe sepsis. Patients with severe sepsis that have continued organ system impairment and/or low blood pressure that does not respond to treatment with adequate fluid replacement are considered to be in septic shock. Severe sepsis and septic shock impacts approximately 50,000 patients in NY each year, and on average almost 30% patients will die from this syndrome. In addition, many more may experience lifelong impairments as a result the broad impact that sepsis may have on organ and tissue function. For purposes this report, the term sepsis will be used to indicate severe sepsis and septic shock. The combination early detection sepsis coupled with appropriate interventions can significantly improve the chances survival for patients with all types sepsis. This public report is one part a statewide initiative to reduce the impact this deadly condition by improving early detection and intervention for patients with sepsis, focused on the most deadly form severe sepsis and septic shock. The New York State Sepsis Initiative This report describes outcomes for patients with severe sepsis/septic shock being treated in hospitals across New York during 2015. It is the first its kind in the nation. It represents considerable efforts by New York State hospitals and clinicians, over the past three years, to measure and improve care for individuals with this common, complex, and lethal, condition. Beginning in 2014 each acute care hospital in New York that provides care to patients with sepsis was required by amendment Title 10 the New York State Codes, Rules and Regulations (Sections 405.2 and 405.4) to develop and implement evidence informed sepsis protocols which describe their approach to both early recognition and treatment sepsis patients. In addition, hospitals were required to report data to the New York State Department Health (Department) beginning in 2014 that are used to calculate each hospital s performance on key measures early treatment and protocol use. s were also required to submit sufficient clinical information on each patient with sepsis to allow the Department to develop a methodology to evaluate risk adjusted mortality rates for each hospital. Risk adjustment permits comparison hospital performance and takes into consideration the different mix demographic and comorbidity attributes, including sepsis severity, patients cared for within each hospital. What follows is the report these results for use protocols, adherence to key interventions within those protocols within specific recommended time frames and risk adjusted mortality rates (adults) for each reporting hospital in New York. Public reporting hospital performance is one dimension New York s overall initiative to focus quality and safety improvement efforts on the identification and care patients with sepsis in New York. 1
Percentage New York State Report on Sepsis Care Improvement Initiative: Quality Performance Statewide Trends Data reported by hospitals for 10 quarters (second quarter 2014 through third quarter 2016) are the basis for the following trend analysis. Despite the early nature this initiative we can demonstrate encouraging improvements in protocol initiation, rapid and early treatment, and mortality over time. The graphs below show statewide changes over each quarter beginning in 2014 through the most recent quarter data in 2016. NOTE: For the time determined measures in the trend graphs for this section, time zero is defined as the date/time when each hospital determined that its protocol had been initiated for each patient. Protocol Initiation Figure 1 shows the percentage adult patients (age 18) with severe sepsis or septic shock for whom a protocol was initiated at the treating hospital. At the onset the initiative, the protocol was initiated for 73.7% patients. This percentage has increased progressively to a high 84.7% in quarter three 2016. 100% 90% 80% 70% 73.7% 75.9% 77.4% 79.1% 81.4% 81.0% 83.1% 81.7% 83.8% 84.7% 60% 50% 40% 30% 20% 10% 0% 2014Q2 2014Q3 2014Q4 2015Q1 2015Q2 2015Q3 2015Q4 2016Q1 2016Q2 2016Q3 Quarter Figure 1. Adult Protocol Initiation: Quarter Two, 2014 through Quarter Three, 2016* (*) excludes patients with clinical contraindications for protocol interventions or who died within six hours Figure 2 shows the percentage pediatric patients (age < 18) with severe sepsis or septic shock for whom a protocol was initiated at the treating hospital. At the onset the initiative, the protocol was initiated for 80.6% patients. This percentage fluctuated in subsequent quarters possibly due to the low number pediatric sepsis cases across the state. In the most recent quarter 2016, a protocol was initiated at the treating hospital for 85.3% the pediatric patients with severe sepsis or septic shock. 2
Percentage New York State Report on Sepsis Care Improvement Initiative: Quality Performance 100% 90% 80% 70% 80.6% 82.4% 83.8% 76.1% 78.0% 68.7% 77.5% 80.8% 87.6% 85.3% 60% 50% 40% 30% 20% 10% 0% 2014Q2 2014Q3 2014Q4 2015Q1 2015Q2 2015Q3 2015Q4 2016Q1 2016Q2 2016Q3 Quarter Figure 2. Pediatric Protocol Initiation: Quarter Two, 2014 through Quarter Three, 2016* (*) excludes patients with clinical contraindications for protocol interventions or who died within one hour Rapid and Early Treatment Figure 3 shows the percentage adult patients (age 18) with severe sepsis or septic shock for whom all the recommended early treatments in the 3-hour early management bundle were administered. At the onset the initiative, the treatments in the 3-hour early management bundle were successfully completed for 41.5% those patients with severe sepsis or septic shock for whom a protocol was initiated. This percentage increased progressively and reached a high 55.2% in quarter three 2016. Figure four shows the percentage adult patients (age 18) with severe sepsis or septic shock for whom all the recommended early treatments in the 6-hour early management bundle were administered. At the onset the initiative, the treatments in the 6-hour bundle measure were successfully completed for 22.6% those patients with severe sepsis or septic shock for whom a protocol was initiated. This percentage increased progressively and reached a high 36.4% in quarter three 2016. 3
Percentage Percentage New York State Report on Sepsis Care Improvement Initiative: Quality Performance 100% 90% 80% 70% 60% 50% 40% 41.5% 44.7% 47.4% 48.5% 50.4% 50.4% 54.1% 52.6% 55.2% 55.2% 30% 20% 10% 0% 2014Q2 2014Q3 2014Q4 2015Q1 2015Q2 2015Q3 2015Q4 2016Q1 2016Q2 2016Q3 Quarter Figure 3. Adult Early Intervention (3-Hour Early Management Bundle): Quarter Two, 2014 through Quarter Three, 2016* (*) excludes patients with clinical contraindications for protocol interventions or who died within six hours 100% 90% 80% 70% 60% 50% 40% 30% 22.6% 24.8% 26.4% 28.0% 30.5% 30.6% 33.1% 33.9% 35.2% 36.4% 20% 10% 0% 2014Q2 2014Q3 2014Q4 2015Q1 2015Q2 2015Q3 2015Q4 2016Q1 2016Q2 2016Q3 Quarter Figure 4. Adult Early Intervention (6-Hour Early Management Bundle): Quarter Two, 2014 through Quarter Three, 2016* (*) excludes patients with clinical contraindications for protocol interventions or who died within six hours Figure 5 shows the percentage pediatric patients (age < 18) with severe sepsis or septic shock for whom all the recommended early treatments in the 1-hour early management bundle were administered. At the onset the initiative, the treatments in the 1-hour bundle 4
Percentage New York State Report on Sepsis Care Improvement Initiative: Quality Performance measure were successfully completed for 4.9% those patients with severe sepsis or septic shock for whom a protocol was initiated. This percentage was higher in subsequent quarters but demonstrated significant unexplained fluctuations potentially related to small case volume. In the most recent quarter, quarter three 2016, the treatments in the 1-hour bundle were successfully administered to 17.6% those pediatric patients diagnosed with severe sepsis or septic shock who had a protocol initiated at the treating hospital. 50% 45% 40% 35% 30% 25% 22.0% 19.9% 28.3% 23.1% 24.3% 20% 16.7% 17.6% 15% 13.3% 12.7% 10% 5% 0% 4.9% 2014Q2 2014Q3 2014Q4 2015Q1 2015Q2 2015Q3 2015Q4 2016Q1 2016Q2 2016Q3 Quarter Figure 5. Pediatric Early Intervention (1-Hour Early Management Bundle): Quarter Two, 2014 through Quarter Three, 2016* (*) excludes patients with clinical contraindications for protocol interventions or who died within one hour Mortality Figure six shows the percentage adult patients (age 18) with severe sepsis or septic shock who died during their hospital stay. At the onset the initiative, approximately 30.2% those patients treated for severe sepsis or septic shock died in the hospital. This percentage decreased over time and reached a low 25.4% in quarter three 2016. Figure seven shows the percentage pediatric patients (age < 18) with severe sepsis or septic shock who died during their hospital stay. At the onset the initiative, approximately 6.8% those patients treated for severe sepsis or septic shock died in the hospital. This percentage fluctuated over time reaching a high 15.3% in quarter one 2015 and a low 6.5% in quarter 3 2015. 5
Percentage Percentage New York State Report on Sepsis Care Improvement Initiative: Quality Performance 100.0% 90.0% 80.0% 70.0% 60.0% 50.0% 40.0% 30.0% 30.2% 30.4% 30.6% 32.0% 28.4% 27.9% 27.3% 28.0% 26.4% 25.4% 20.0% 10.0% 0.0% 2014Q2 2014Q3 2014Q4 2015Q1 2015Q2 2015Q3 2015Q4 2016Q1 2016Q2 2016Q3 Quarter Figure 6. Adult In- Mortality: Quarter Two, 2014 through Quarter Three, 2016 50.0% 45.0% 40.0% 35.0% 30.0% 25.0% 20.0% 15.0% 10.0% 6.8% 10.0% 11.5% 15.3% 8.8% 6.5% 8.6% 9.5% 6.8% 10.5% 5.0% 0.0% 2014Q2 2014Q3 2014Q4 2015Q1 2015Q2 2015Q3 2015Q4 2016Q1 2016Q2 2016Q3 Quarter Figure 7. Pediatric In- Mortality: Quarter Two, 2014 through Quarter Three, 2016 6
Sepsis Improvement Initiatives: Collaborations The Department is collaborating with federal, state, and private initiatives to improve sepsis awareness, advance sepsis care, and make maximal use the data collected from hospitals to better understand which clinical practices are influencing survival and other important outcomes for patients. Several these efforts to improve sepsis care are described below. Sepsis Advisory Group The Department convenes an ad hoc group clinicians from across New York that has assisted with the development and implementation the initiative since 2013. This diverse expert group includes both adult and pediatric specialists who treat patients with sepsis. The advisory group has provided key input into the structure on-going quarterly performance reports presented to each hospital on their protocol use, protocol adherence, and mortality results compared to statewide averages as well as trended over time. These interim feedback reports have provided the stimulus for hospitals leading to the improvements we have seen over time. In addition to providing input in the refinement our data collection and measurement process, the group will advise the department on new developments and interventions for patients with sepsis, including treatments and processes care delivery, that show promise to improve outcomes for patients with sepsis throughout New York. With the completion the first quality reporting cycle for hospitals the advisory group will transition from measurement and data development to active use results aimed at identifying and disseminating promising clinical interventions and system improvements from those hospitals with exceptional results. IPRO, Implementation Business Partner IPRO assisted the Department throughout the initiative from the review hospital sepsis protocols to development data dictionary, feedback reports, and analyses. Key activities included the streamlining electronic data collection, ensuring data integrity, customizing reports, providing webinars, and helpdesk support to hospitals. Partnership For Patients (P4P) The Center for Medicare and Medicaid Services (CMS) has awarded the hospital associations in New York State with grants to support a variety quality and safety improvements focused on inpatient care. The Healthcare Association New York State (HANYS) and the Greater New York Association (GNYHA) have worked in collaboration with the participating hospitals and the Department in making sepsis care one the priorities for this improvement work. This initiative aims to help hospitals improve sepsis care processes by supporting front line staff adherence to their protocols. The Department, in collaboration with IPRO, ensures that this innovative initiative can make maximal use the data that has been collected to date in order to focus improvement activities on key clinical interventions that create challenges for clinicians and hospitals and be able to share promising practices. 7
For example, the P4P has focused on improvement in the following areas: Early Identification Patients with Sepsis o Implementation screening processes leveraging electronic health records to create early warning systems Timely Treatment o Revising workflow processes including point care testing, rapid antibiotic access, decision supported order sets, and Rapid Response or Code Sepsis teams Clinical Management o Standardized tools for communication between hospitals and clinicians Case Review o Real time review cases with identification educational opportunities for clinicians IPRO/CMS Collaboration: Community Based Providers The Centers for Medicare and Medicaid Services (CMS) awarded the Atlantic Quality Innovation Network led by IPRO a two-year contract award to provide education to improve early identification, treatment and management sepsis patients among pre-hospital providers and the general public in the Northeastern and Central New York region and the Charleston area in South Carolina. The ultimate objective this initiative is to reduce sepsis-related morbidity and mortality through education utilizing evidence based practices and protocols. Targeted community-based healthcare providers include skilled nursing facilities, home health agencies and physician practices. Using a Train-the-Trainer model, more than 7,500 clinical and nonclinical staff have been trained on identifying the early signs and symptoms sepsis as well as recognizing sepsis as a medical emergency. Collaborative partnerships with the Center for Disease Control & Prevention, Home Care Association New York State, Sepsis Alliance, and The Rory Staunton Foundation have garnered wide spread exposure the initiative outside the target regions. Process, proximal and outcome performance measures are used to evaluate the effectiveness the Community Based Sepsis Initiative. Private Foundations Several private foundations have provided support and assistance in raising public awareness regarding sepsis which has amplified the work the initiative in New York. In addition, the Rory Staunton Foundation created by the Staunton family and named for Rory Staunton, a 12 year old New York resident who died from sepsis in 2012 was instrumental in advocating for the existing regulations ( Rory s Regulations ) in New York and now, in other states as well. Other organizations, such as the Sepsis Alliance, have also played an important national role in bringing attention and focus to sepsis care. 8
Measure Descriptions The following measures are included in this report and are briefly summarized below. The measures evaluate several key processes care (and one important outcome mortality) that can increase the probability surviving an episode sepsis. The Adult New York State sepsis process care measures were developed using a National Quality Forum (NQF) measure for guidance: NQF #500 Severe Sepsis and Septic Shock: Management Bundle. The percentage patients who received care using the hospital developed sepsis protocol The percentage adult patients with sepsis treated in the emergency room with the hospital s sepsis protocol who received all the recommended early treatments in the 3- hour early management bundle within three (3) hours their arrival The percentage adult patients with septic shock treated in the emergency room with the hospital s sepsis protocol who received all the recommended early treatments in the 6-hour early management bundle within six (6) hours their arrival The percentage pediatric patients with sepsis treated in the emergency room with the hospital s sepsis protocol who received all the recommended early treatments within one (1) hour their arrival The risk adjusted inpatient mortality (death) rate adult patients in each hospital Percentage Receiving Protocol Treatment (Adult and Pediatric) This reported rate for each hospital describes what percentage sepsis patients in each hospital received care consistent with the initiation their formal protocol, excluding those cases with identified (and justified) clinical or advanced directive exceptions. After adjusting for patient factors, the department s analysis the data shows that the odds dying are 21% less for adult patients who receive protocol driven treatments compared to patients who do not receive protocol driven treatments. While all patients with sepsis are required to be considered for protocol treatment, hospitals and clinicians may have valid reasons for not making use a protocol. These reasons include that the patient has advanced directives in place restricting the use some/all protocol interventions, patient/family declines interventions, clinical contraindications for some/all the protocol interventions, or patient is enrolled in a research study involving different interventions or approaches. s were permitted to exclude those specific patients from protocol measures and overall approximately 3% pediatric patients and 7% adult cases were excluded from protocol interventions. The majority these exclusions were due to clinical contraindications. However, there are other instances in which a patient may have not received protocol driven care unrelated to these exceptions, which may describe missed opportunities in the identification sepsis or the delayed application protocol interventions. We asked hospitals to indicate for each case whether their protocol was initiated (using their own definition protocol initiation described to the Department when they submitted their protocols for review and approval) as well as indicate whether there were exclusions as described above. 9
Percentage Receiving All Treatments in Three Hours (Adult) For purposes evaluating the ability hospitals to implement their own protocols, we used an existing measure (with minor modifications) that has been approved by the National Quality Forum (NQF) for sepsis treatment. NQF is a public/private partnership organization that reviews quality measures for use in public reporting and payment programs for the Center for Medicare and Medicaid Services. This composite measure (NQF #500 Severe Sepsis and Septic Shock: Management Bundle) includes a three (3) hour timed bundled measure in which the beginning time (or time zero ) for patients with severe sepsis and septic shock presenting in the Emergency Department is defined as the recorded triage time. The interventions within this measure include measurement a blood lactate level, obtaining blood cultures prior to giving antibiotics, and administering broad spectrum antibiotics. These interventions collectively have been demonstrated to help direct appropriate care (lactate and blood cultures), as well as provide early important treatment (antibiotics) for life threatening infections. Patients with clinical exclusions and patients who have been transferred from or to another acute care hospital ARE EXCLUDED from this measure. After adjusting for patient factors, the Department s analysis the data shows that the odds dying are 27% less for adult patients who receive all the recommended treatments within three hours compared to patients who do not receive all the recommended treatments. Percentage With Septic Shock Receiving All Treatments in Six Hours (Adult) Patients with septic shock require additional treatments to stabilize and treat their condition beyond the interventions described in the three hour bundle. These patients have very low blood pressure or significantly elevated blood lactate levels that suggest a more serious condition. For those patients there are three additional interventions that comprise the six hour bundle. These interventions address supporting blood pressure and organ function with both fluids and other medications (vasopressors) as well as re-measuring blood lactate levels when the initial lactate is abnormal. This measure, using the same time zero as the three hour bundle, measures the percentage patients with septic shock (a subset all patients) that received all the three hour bundle as well as the three additional interventions described in this section. Patients with clinical exclusions and patients who have been transferred from or to another acute care hospital ARE EXCLUDED from this measure. After adjusting for patient factors, the Department s analysis the data shows that the odds dying are 26% less for adult patients who receive all the recommended treatments within six hours compared to patients who do not receive all the recommended treatments. Percentage Receiving All Treatments Within One Hour (Pediatric) As with adults, early treatment with fluids and antibiotics to children with severe sepsis is associated with improved survival. Using guideline recommendations from the Pediatric Advanced Life Support (PALS) program the American Heart Association this measure evaluates the percentage pediatric patients with sepsis that received parenteral fluids, blood cultures, and antibiotics within one hour their presentation in the emergency room. Patients with clinical exclusions and patients who have been transferred from or to another acute care hospital ARE EXCLUDED from this measure. 10
Risk Adjusted Mortality Rates (Adult) The use sepsis protocols and the measures to evaluate protocol adherence are important to patients in so far as they can improve the probability survival. In order to fairly compare hospitals on this critical outcome (survival) it is first necessary to be able to account for patient differences which can increase the risk dying from sepsis. A risk adjustment is used which takes into account accompanying chronic illnesses which can complicate treatment and outcomes for patients with sepsis, patient demographic factors such as age, and the severity sepsis for each patient. This measure describes the risk adjusted percentage all patients with sepsis at each hospital who died during that hospital stay. This measure excludes acute care transfer patients, patients with advanced directives that restricted the use any protocol interventions, or patients that refused any the protocol interventions. More detail regarding the risk adjustment methodology can be found in Technical Appendix A. Performance Data The clinical sepsis data submitted by the hospitals for patients with severe sepsis and septic shock for calendar year 2015 was used to calculate the sepsis performance metrics listed above. The performance measures are only reported for those hospitals with greater than 10 sepsis cases in calendar year 2015. After calculating the performance measures for each hospital, the data for each individual measure was ordered from the lowest percentage to the highest percentage and divided into quintiles. Each hospital was assigned to a performance level category based on the quintile into which their percentage fell for a given measure. Those hospitals ranked in quintile 1 are the lowest performers and those hospitals ranked in quintile 5 are the highest performers. Table one shows the quintiles, category assignment, and the percentages assigned to each category for the three adult measures protocol initiated, 3-hour bundle, and 6-hour bundle. Table one shows, for example, that a hospital was ranked into quintile 1 for the protocol initiated measure if the protocol was initiated for between 4 and 70.23% the sepsis cases treated at the hospital. The value 70.23% was chosen as the cutf for quintile 1 because 20% the values for this measure were at or below 70.23% in calendar year 2015. Table 1. Category Assignment for the Adult Sepsis Performance Measures Category (Performance Level) Summary Table Symbol Ranking Percentiles Protocol Initiated (%) 3-Hour Bundle (%) 6-Hour Bundle (%) Best 5 Highest 80 th 100 th 99.95-100.00 69.24-88.46 41.68-75.00 4 High 60 th 80 th 93.87-99.94 60.18-69.23 33.80-41.67 3 Middle 40 th 80 th 85.72-93.86 52.90-60.17 27.15-33.79 2 Low 20 th 40 th 70.24-85.71 45.46-52.89 20.37-27.14 1 Lowest 0 th 20 th 4.00-70.23 7.27-45.45 0.00-20.36 Worst 11
Table two shows each hospital along with the performance level category for the adult performance measures using symbols that correspond to the category symbols shown in Table 1. The data is suppressed due to small sample size (S.S.) when a hospital did not have more than 10 patients for the measure. An N.C. indicates that the measure was not calculated because the hospital did not have any patients that satisfied the criteria for inclusion in the measure. The detailed data with specific hospital and statewide values for the individual performance measures is shown in Technical Appendix B. Table 2. Adult Sepsis Performance Measure Summary Report by Protocol Initiated (Adult) 3-Hour Bundle (Adult) 6-Hour Bundle (Adult) Adirondack Medical Center-Saranac Lake Site S.S. S.S. Albany Medical Center Albany Memorial Alice Hyde Medical Center Arnot Ogden Medical Center Auburn Community Aurelia Osborn Fox Memorial S.S. S.S. Bellevue Center Bertrand Chaffee S.S. S.S. Bon Secours Community Bronx-Lebanon Ctr - Concourse Div. Best Highest High Middle Low Lowest Brookdale Medical Center Brookhaven Memorial Medical Center Worst Brooklyn Center - Downtown Campus Brooks Memorial S.S. S.S. Buffalo General Medical Center Canton-Potsdam Catskill Regional Medical Center Cayuga Medical Center at Ithaca Champlain Valley Physicians Med Ctr Chenango Memorial Claxton-Hepburn Medical Center Clifton Springs and Clinic S.S. S.S. 12
Protocol Initiated (Adult) 3-Hour Bundle (Adult) 6-Hour Bundle (Adult) Columbia Memorial Community Memorial S.S. S.S. Coney Island Corning Cortland Regional Medical Center Crouse Degraff Memorial Eastern Long Island S.S. S.S. Eastern Niagara - Lockport Division Elizabethtown Community S.S. S.S. Ellis Elmhurst Center Erie County Medical Center F F Thompson Faxton-St Lukes Healthcare St Lukes Division Flushing Medical Center Best Highest High Middle Low Lowest Forest Hills Franklin Worst Geneva General Glen Cove Glens Falls Good Samaritan Medical Center Good Samaritan Suffern Gouverneur S.S. S.S. Harlem Center HealthAlliance Broadway Campus Highland Hudson Valley Center Huntington 13
Protocol Initiated (Adult) 3-Hour Bundle (Adult) 6-Hour Bundle (Adult) Interfaith Medical Center Jacobi Medical Center Jamaica Medical Center John T Mather Memorial Jones Memorial Kenmore Mercy Kings County Center Kingsbrook Jewish Medical Center Lawrence Center Lenox Hill Lewis County General S.S. S.S. Lincoln Medical & Mental Health Center Little Falls S.S. S.S. Long Island Jewish Medical Center Best Highest High Middle Low Lutheran Medical Center Maimonides Medical Center Lowest Mary Imogene Bassett Massena Memorial Worst Medina Memorial Health Care System Memorial for Cancer and Allied Dis Mercy Mercy Medical Center Metropolitan Center Millard Fillmore Suburban Montefiore Med Center - Einstein College Div Montefiore Medical Center - Moses Div Montefiore Medical Center-Wakefield Montefiore Mount Vernon Montefiore New Rochelle 14
Protocol Initiated (Adult) 3-Hour Bundle (Adult) 6-Hour Bundle (Adult) Moses-Ludington S.S. S.S. Mount Sinai Beth Israel Mount Sinai Beth Israel Brooklyn Mount Sinai Mount Sinai - Queens Mount Sinai Roosevelt Mount Sinai St. Lukes Best Highest Mount St Marys and Health Center Nassau University Medical Center Nathan Littauer New York Community Brooklyn, Inc New York Medical Center Queens New York Methodist New York Presbyterian - Allen New York Presbyterian - Columbia New York Presbyterian - Weill Cornell High Middle Low Lowest New York-Presbyterian/Lower Manhattan Newark-Wayne Community Worst Niagara Falls Memorial Medical Center Nicholas H Noyes Memorial S.S. S.S. North Central Bronx North Shore University Northern Dutchess Northern Westchester Nyack NYU s Center OConnor N.C. N.C. Olean General Oneida Healthcare 15
Protocol Initiated (Adult) 3-Hour Bundle (Adult) 6-Hour Bundle (Adult) Orange Regional Medical Ctr-Goshen Campus Oswego Our Lady Lourdes Memorial Inc Peconic Bay Medical Center Phelps Memorial Assn Plainview Putnam Center Queens Center Richmond University Medical Center River, Inc. S.S. S.S. Rochester General Rome Memorial, Inc Roswell Park Cancer Institute N.C. N.C. Samaritan Samaritan Medical Center Saratoga Best Highest High Middle Low Lowest SBH Health System Sisters Charity Worst Sisters Charity - St Joseph Campus SJRH - Andrus Pavilion Soldiers and Sailors Memorial S.S. South Nassau Communities Southampton Southside St Anthony Community St Catherine Siena St Charles St Elizabeth Medical Center St Francis 16
Protocol Initiated (Adult) 3-Hour Bundle (Adult) 6-Hour Bundle (Adult) St Francis - Poughkeepsie St James Mercy S.S. S.S. St Johns Episcopal So Shore St Josephs Health Center St Josephs Medical Center St Lukes Cornwall /Newburgh St Peters St. Joseph St. Marys Healthcare St. Marys Staten Island University - North Staten Island University - South Strong Memorial Syosset The Unity Rochester Tri Town Regional Healthcare S.S. S.S. United Health Services s Inc. - Binghamton General United Health Services s Inc. - Wilson Medical Center United Memorial Medical Center North Street Campus University (Stonybrook) Best Highest High Middle Low Lowest Worst University Brooklyn University SUNY Health Science Center (Syracuse) Upstate University at Community General Vassar Brothers Medical Center Westchester Medical Center White Plains Center Winthrop-University Womans Christian Association Woodhull Medical & Mental Health Center 17
Protocol Initiated (Adult) 3-Hour Bundle (Adult) 6-Hour Bundle (Adult) Wyckf Heights Medical Center Wyoming County Community S.S. S.S. Table three shows the high and low performing hospitals for the adult risk-adjusted mortality rate outcome measure. High and low performing hospitals were determined based on whether there was a statistically significant difference between the expected number deaths based on the statistical model and the observed number deaths based on the submitted data. Those with a statistically significant difference between the expected and observed number deaths where the observed number deaths was less than expected were assigned to the high performer category ( ). Those with a statistically significant difference between the expected and observed number deaths where the observed number deaths was higher than expected were assigned to the low performer category ( ). Those without a statistically significant difference between the expected and observed number deaths were assigned to the middle performer category ( ). The data is suppressed due to small sample size (S.S.) when a hospital did not have more than 10 patients for the measure. An N.C. indicates that the measure was not calculated because the hospital did not have enough patients to calculate a valid risk adjusted mortality rate. The detailed data with specific hospital values for the riskadjusted mortality rate outcome measure is shown in Technical Appendix B. Table 3. Adult Sepsis Outcome Measure Summary Report by Adirondack Medical Center- Saranac Lake Site Albany Medical Center Albany Memorial Risk Adjusted Mortality Rate Brooklyn Center - Downtown Campus Brooks Memorial Buffalo General Medical Center Risk Adjusted Mortality Rate Best High Alice Hyde Medical Center Arnot Ogden Medical Center Auburn Community Aurelia Osborn Fox Memorial Bellevue Center Bertrand Chaffee Bon Secours Community Bronx-Lebanon Ctr - Concourse Div. Brookdale Medical Center Brookhaven Memorial Medical Center Canton-Potsdam Catskill Regional Medical Center Cayuga Medical Center at Ithaca Champlain Valley Physicians Med Ctr Chenango Memorial Claxton-Hepburn Medical Center Clifton Springs and Clinic Columbia Memorial Community Memorial Coney Island Middle Low Worst 18
Risk Adjusted Mortality Rate Risk Adjusted Mortality Rate Corning Jamaica Medical Center Cortland Regional Medical Center John T Mather Memorial Crouse Jones Memorial Degraff Memorial Kenmore Mercy Eastern Long Island Eastern Niagara - Lockport Division Elizabethtown Community Ellis Elmhurst Center Erie County Medical Center F F Thompson Faxton-St Lukes Healthcare St Lukes Division Flushing Medical Center Forest Hills Franklin Geneva General Glen Cove Glens Falls Good Samaritan Medical Center Good Samaritan Suffern Gouverneur Harlem Center HealthAlliance Broadway Campus Highland Hudson Valley Center Huntington Interfaith Medical Center Jacobi Medical Center Kings County Center Kingsbrook Jewish Medical Center Lawrence Center Lenox Hill Lewis County General Lincoln Medical & Mental Health Center Little Falls Long Island Jewish Medical Center Lutheran Medical Center Maimonides Medical Center Mary Imogene Bassett Massena Memorial Medina Memorial Health Care System Memorial for Cancer and Allied Dis Mercy Mercy Medical Center Metropolitan Center Millard Fillmore Suburban Montefiore Med Center - Einstein College Div Montefiore Medical Center - Moses Div Montefiore Medical Center- Wakefield Montefiore Mount Vernon Montefiore New Rochelle Moses-Ludington Best High Middle Low Worst 19
Mount Sinai Beth Israel Risk Adjusted Mortality Rate Orange Regional Medical Ctr- Goshen Campus Risk Adjusted Mortality Rate Mount Sinai Beth Israel Brooklyn Oswego Mount Sinai Our Lady Lourdes Memorial Inc Mount Sinai - Queens Peconic Bay Medical Center Mount Sinai Roosevelt Phelps Memorial Assn Mount Sinai St. Lukes Mount St Marys and Health Center Nassau University Medical Center Nathan Littauer New York Community Brooklyn, Inc New York Medical Center Queens New York Methodist New York Presbyterian - Allen New York Presbyterian - Columbia New York Presbyterian - Weill Cornell New York-Presbyterian/Lower Manhattan Newark-Wayne Community Niagara Falls Memorial Medical Center Nicholas H Noyes Memorial North Central Bronx North Shore University Northern Dutchess Northern Westchester Nyack NYU s Center OConnor Olean General Oneida Healthcare Plainview Putnam Center Queens Center Richmond University Medical Center River, Inc. Rochester General Rome Memorial, Inc Roswell Park Cancer Institute Samaritan Samaritan Medical Center Saratoga SBH Health System Sisters Charity Sisters Charity St. Joseph Campus SJRH - Andrus Pavilion Soldiers and Sailors Memorial South Nassau Communities Southampton Southside St Anthony Community St Catherine Siena St Charles St Elizabeth Medical Center Best High Middle Low Worst 20
Risk Adjusted Mortality Rate Risk Adjusted Mortality Rate St Francis Tri Town Regional Healthcare St Francis - Poughkeepsie St James Mercy St Johns Episcopal So Shore St Josephs Health Center United Health Services s Inc. - Binghamton General United Health Services s Inc. - Wilson Medical Center United Memorial Medical Center North Street Campus University (Stonybrook) Best High St Josephs Medical Center St Lukes Cornwall /Newburgh St Peters St. Joseph University Brooklyn University SUNY Health Science Center Upstate University at Community General Vassar Brothers Medical Center Middle Low St. Marys Healthcare Westchester Medical Center St. Marys Staten Island University - North Staten Island University - South Strong Memorial White Plains Center Winthrop-University Womans Christian Association Woodhull Medical & Mental Health Center Worst Syosset Wyckf Heights Medical Center The Unity Rochester Wyoming County Community Table four shows the quintiles, category assignment, and the percentages assigned to each category for the two pediatric measures protocol initiation and 1-hour bundle. For the protocol initiation measure, all hospitals with a protocol treatment measure percentage 100% were assigned to quintile 5 so there are highest but not high performers for this measure. Risk adjusted mortality rates were not calculated for the pediatric population due to the significantly smaller volume cases for each hospital compared to adult cases and the current lack a standardized, validated risk adjustment model for the pediatric sepsis population. Table five shows each hospital along with the performance level category for the pediatric performance measures using symbols that correspond to the performance level category symbols shown in Table 3. The data is suppressed for small sample sizes (S.S.) when a hospital did not have more than 10 patients for the measure. An N.C. indicates that the measure was not calculated because the hospital did not have any patients that satisfied the criteria for inclusion in the measure. The detailed data with specific hospital and statewide values for the individual performance measures is shown in Technical Appendix B. 21
Table 4. Category Assignment for the Pediatric Sepsis Performance Measures Category (Performance Level ) Summary Table Symbol Percentiles Included Protocol Initiation (%) 1-Hour Bundle (%) Best 5 Highest 80 th 100 th 100.00 21.05 4 High 60 th 80 th 100.00 12.50 21.04 3 Middle 40 th 80 th 93.0 99.99 11.11 12.49 2 Low 20 th 40 th 56.5 92.9 3.92 11.10 1 Lowest 0 th 20 th 15.7 56.4 0.00 3.91 Worst Table 5. Pediatric Sepsis Performance Measures Summary Report by Albany Medical Center Women and Children s Buffalo University Protocol Initiated (Pediatric) 1-Hour Bundle (Pediatric) S.S. S.S. Best Highest Strong Memorial Winthrop-University University SUNY Health Science Center Westchester Medical Center Montefiore Medical Center Henry and Lucy Moses Division Bronx-Lebanon Center Concourse Division Kings County Center Maimonides Medical Center Mount Sinai New York Presbyterian New York Weill Cornell Center NYU s Center New York Presbyterian Columbia Presbyterian Center Long Island Jewish Schneiders Children s Division S.S. S.S. S.S. S.S. S.S. N.C. S.S. High Middle Low Lowest Worst 22
Next Steps The Department has several important initiatives to refine and enhance the utility this data for improvement. Identification and Sharing Promising Practices Being able to identify those facilities with lower sepsis mortality rates now enables the Department and the Sepsis Advisory Group to better explore and identify the specific clinical practices and delivery systems implemented that are likely to be key success elements in improving outcomes. These include, but are not limited to, innovative approaches to early identification high risk patients, rapid response early interventions, mobilization clinical, laboratory, and pharmacy resources within the institution, sepsis protocol content, quality improvement activities, use clinical decision support through electronic medical records, workforce sepsis training and education, and more. Both the Advisory Group and the P4P provide a forum for discussion and dissemination these findings. Data Collection Improvement and Alignment The data dictionary will continue to be streamlined and improved so that hospitals and their data collection staff can completely and accurately report all data elements needed for valid and reliable quality measurement. On-going data audits provide information to both hospitals and to the Department that serve to identify variables requiring further elaboration. During the time this statewide initiative, the Center for Medicare and Medicaid Services (CMS) began to require data collection from hospitals on a sample adult patients with sepsis ( SEP-1 measure). The Department will be working with CMS and with hospitals to align our data collection initiative, where appropriate, with this new federal requirement. Future Measurement: Pediatrics and Morbidity The Department plans to explore the ability to evaluate other important pediatric sepsis outcomes beyond the one hour bundle including risk adjusted mortality. While the number pediatric cases for each hospital will not permit statistically valid comparisons these results could be used for quality improvement and internal hospital benchmarking. Last, there are other outcomes in addition to survival that are important to clinicians and patients related to serious and long lasting organ or tissue damage that can result from sepsis. While there are currently no standardized metrics or data to capture this information the Department is committed to exploring ways to develop new and innovative measures in this important area. 23
Page Intentionally Left Blank 24
Technical Appendix A: Risk Adjustment Methodology The objective the risk adjustment process is to assess hospital performance after accounting for differences in patient case mix between hospitals. In the first part the process, a mortality model estimates the probability in-hospital mortality for each patient with sepsis. This estimate is based on patient demographic, comorbidity, and severity illness characteristics. Multivariable logistic regression was used to determine which variables are important and accurate in estimating the probability mortality for each patient. Treatment variables, within the control or influence the providers and hospital, are not included in the prediction model. Table A1 shows the thirteen (13) variables in this model that are used to estimate the probability mortality as well as the overall performance the model. Variables are primarily binary or categorical; three them are continuous. These are age (in years), first serum lactate (mmol/l), and comorbidity count. The model includes both main effects and interactions. Variables were included as main effects if their p-values were <0.05; interactions had to be statistically significant and clinically relevant. The dataset for model development included 43,204 patients. 10% these patients were randomly selected and set aside to validate the model developed on the other 90% the patients. The risk adjusted model in this report makes use the most recent complete and audited data from four quarters patient data submission in 2015. All patients who were discharged and transferred from one hospital to another were excluded from model development and the application the model to each hospital s result. Patients with advanced care directives in place prior to the episode sepsis who declined sepsis protocol interventions, or who refused sepsis protocol interventions at the time presentation, were removed from the data set. Patients admitted more than once in 2015 for sepsis are represented only once for purposes development the risk adjusted model (using their last admission only). For purposes evaluating each hospital s performance, each admission is included. To assess hospital performance, the probability hospital mortality is calculated for every patient from that hospital using the logistic regression model. These probabilities are summed over all the patients at that hospital to calculate the expected number deaths for that hospital. The actual number deaths is determined for all patients in that hospital as well. The standardized mortality ratio (SMR) is calculated by dividing the observed by the expected number deaths in each hospital. The SMR was then multiplied by the statewide mortality rate to obtain a risk adjusted mortality rate (RAMR) and a 95% confidence interval for the RAMR. The RAMR provides the best estimate what each hospital s mortality rate would have been if the hospital had a case mix that was identical to the statewide mix. If the confidence interval for a hospital s RAMR is entirely below the statewide rate, the hospital performed significantly better than the state average. If the hospital s confidence interval was entirely above the statewide rate the hospital performed significantly worse than the statewide rate. Figure A1 contains a plot showing the RAMR and confidence interval for each hospital. The highest performing hospitals are displayed in blue and the lowest performing hospitals are displayed in gold. 25
Table A1. Variables in the Risk Adjusted Mortality Rate (RAMR) model Main effects or Interactions % β Adjusted OR p-value Race/Ethnicity main effect White, Non-Hispanic 56.7 -Reference- Black, Non-Hispanic 15.3 0.188 1.21 < 0.001 Hispanic 9.2-0.073 0.93 0.112 Multi-racial 1.6 0.165 1.18 0.096 Unknown, Non-Hispanic 7.2 0.043 1.04 0.382 Unknown 10.0 0.101 1.11 0.016 Payer - main effect Medicare 59.9 -Reference- Medicaid 16.6 0.099 1.11 0.011 Private HMO 18.9 0.067 1.07 0.056 Self-pay 1.3 0.646 1.91 < 0.001 Other 3.2-0.023 0.98 0.750 Site infection - main effect Urinary 22.6 -Reference- Respiratory 39.9 0.599 1.82 < 0.001 Gastrointestinal 12.6 0.543 1.72 < 0.001 Skin 6.7 0.484 1.62 < 0.001 Central Nervous System 0.6 0.723 2.06 < 0.001 Other 8.8 0.558 1.75 < 0.001 Unknown 8.7 0.900 2.46 < 0.001 Admission source - main effect Non-health facility, POA 68.2 -Reference- Clinic 4.1-0.033 0.97 0.605 Different 6.3 0.418 1.52 < 0.001 SNF/ICF 19.6 0.289 1.34 < 0.001 Another HC facility 0.8 0.278 1.32 0.040 Between unit transfer 0.4 0.459 1.58 0.013 Hospice 0.1 0.727 2.07 0.075 Other 0.5-0.330 0.72 0.106 Lower respiratory infection No 50.4 -Reference- Yes 49.6 0.281 -- < 0.001 MV severity No 87.6 -Reference- Yes 12.4 0.519 -- < 0.001 Lower respiratory infection#mv severity Yes#Yes -- -0.398 0.89 < 0.001 Septic shock diagnosis Severe Sepsis 51.3 -Reference- Septic Shock 48.7 0.770 2.16 < 0.001 Platelet count or Thrombocytopenia No 76.3 -Reference- Yes 23.7 0.285 1.33 < 0.001 Metastatic cancer No 89.3 -Reference- Yes 10.7 0.460 1.58 < 0.001 Lymphoma/Leukemia/Multiple Myeloma No 94.8 -Reference- Yes 5.2 0.135 1.14 0.011 Age 100 0.060 -- < 0.001 Square root comorbidity count -- 2.410 -- < 0.001 Age#square root comorbidity count -- -0.021 Varies < 0.001 Serum Lactate 100 0.245 -- < 0.001 Serum Lactate*Serum Lactate 100-0.002 -- 0.001 Serum Lactate#square root comorbidity count -- -0.043 Varies < 0.001 Intercept = -8.548 C Statistic = 0.773 26
Figure A1. RAMR and 95% Confidence Interval by 27
Page Intentionally Left Blank 28
Technical Appendix B The following tables show data summaries performance and outcome measures for inpatient sepsis care for New York State s. Table B1 contains performance and outcome measures for inpatient adult (age 18) sepsis care for New York State s. This table includes the Risk Adjusted Mortality Rates (RAMR) per 100 sepsis patients along with the following quality measures: protocol initiated, 3- hour bundle, and 6-hour bundle. The RAMR (N1) includes all patients except for those with advanced directives, transfer patients, and patients who declined interventions. See Technical Appendix A for additional information about the RAMR model. The Protocol Initiated measure (N2) includes all patients except for those excluded from the protocol or who died within six hours. This measure indicates the percentage patients in the denominator for whom a protocol was initiated. The protocol can be initiated in the ED, ward, or ICU. This measure is only reported for those hospitals with greater than 10 adult sepsis cases in 2015. The 3-hour and 6-hour bundle measures (N3) include all patients with a protocol initiated in the ED who were eligible for the bundle. Transfer cases were excluded from the bundle measures. These two measures report the percentage bundle-eligible patients with a protocol initiated in the ED who received all the interventions specified in the bundle in the time frame. This measure is only reported for those hospitals with greater than 10 adult sepsis cases in 2015. Table B2 contains the performance measures for inpatient pediatric (age < 18) sepsis care for New York State s. This table includes the following performance measures: protocol initiated and 1-hour bundle. The Protocol Initiated measure (N2) includes all patients except for those excluded from the protocol or those who died within one hour. This measure indicates the percentage patients in the denominator for whom a protocol was initiated. The protocol can be initiated in the ED, ward, or ICU. This measure is only reported for those hospitals with greater than 10 pediatric sepsis cases in 2015. The 1-hour bundle measure (N3) includes all patients with a protocol initiated in the ED who were eligible for the bundle. Transfer cases were excluded. This measure reports the percentage bundle-eligible patients with a protocol initiated in the ED who received all the interventions specified in the bundle in the time frame. This measure is only reported for those hospitals with greater than 10 pediatric sepsis cases in 2015. In both tables, the highest performers are highlighted in blue and the lowest performers are highlighted in gold. The cells that contain an S.S. indicate that the data was suppressed due to low counts. The cells that contain an N.C. indicate that the measure was not calculated because the hospital did not have any patients that satisfied the criteria for inclusion in the measure (performance measures) or the hospital did not have enough patients to calculate a valid risk adjusted mortality rate (outcome measure). 29